Nuclear magnetic resonance study of the influence of aqueous sodium

Jul 8, 1971 - Solution Conformations of Uracil Nucleosides and Nucleotides*. Thomas Schleich.f Barry J. Blackburn, t Roy D. Tapper, § and Ian C. P. S...
0 downloads 0 Views 1MB Size
Biochemistry @ Copyrighf 1972 by rhe American Chemical Society

Volume 11, Number 2

January 18,1972

A Nuclear Magnetic Resonance Study of the Influence of Aqueous Sodium Perchlorate and Temperature on the Solution Conformations of Uracil Nucleosides and Nucleotides” Thomas Schleich,? Barry J. Blackburn,$ Roy D. Lapper,$ and Ian C. P. Smith7

ABSTRACT : Complete high-resolution proton magnetic spectral analyses for uridine, uridine 3’-phosphate, 0-pseudouridine, 0-pseudouridine 3 ’-phosphate, and deoxyuridine indicate that changes in temperature and/or the addition of sodium perchlorate, a potent biopolymer conformation destabilizing salt, result only in minimal changes in the solution conformations. The conformations of the above nucleosides and nucleotides are all characterized by: (a) a sugar-base torsional angle corresponding to an anti conformation; (b) a ribose ring which is a rapid time-average blend of conformers; (c) an exocyclic CH20Hgroup which favors the gauche,gauche rotamer, except for deoxyuridine where the rotamers gauche,gauche and gauche,trdns have approximately equal probabilities; (d) in the 3’ nucleotides, an angle of approximately 50” between H3C3fO3’ and C3’O3’P(the cis conformation is assigned to zero angle) as judged from the phosphorus-hydrogen coupling constants. With the exception of the dynamic

P

olynucleotide structures, like the fibrous and globular proteins, are susceptible to an enhanced ease of thermal unfolding or disruption of the “native” conformation in the presence of neutral salt denaturants in aqueous solution. The salt-induced depressions of the transition temperatures for unfolding of collagen, ribonuclease, and calf thymus DNA are approximately the same (the midpoints of the thermal transition span a 70” range), having a value of approximately

* From the Biochemistry Laboratory, National Research Council of Canada, Ottawa, Canada KIA OR6. Received July 8, 1971. N R C C Publication Number 12347. Visiting National Research Council Summer Scientist, 1970. Permanent address : Division of Natural Sciences, University of California, Santa Cruz, Calif. 95060. This investigation was supported, in part, by Research Grant GB-19503 from the National Science Foundation (to T. S.). 3 NRCC Postdoctoral Fellow, 1968-1970. Present address : Department of Chemistry, University of Winnipeg, Winnipeg, Can. .$NRCC Postdoctoral Fellow, 197Ck1972. 7 To whom to address correspondence.

processes in b, c, and d, the conformations in solution follow the general preferences indicated by X-ray studies of single crystals of similar compounds. No correlation of the small changes in conformation could be established with alterations in water structure due to thermal and solute-induced perturbations. The fact that the nucleoside and nucleotide conformations are insensitive to thermal and solute perturbation suggests that the site for the destabilization of polynucleotide conformation is at the level of base-base stacking and interbase hydrogen-bonding interactions. Furthermore the conclusions of this study and our interpretation of the nmr data of Kreishman and Chan (Biopolymers 10, 159 (1971)) for polyuridylic acid and uridine 5 ‘-phosphate suggest that the relatively large characteristic ratio for polyuridylic acid originates in the rotational restrictions found in the monomers themselves.

10“/mole of added sodium perchlorate. These observations suggest that the salt denaturation phenomenon transcends the chemical composition details of a particular biopolymer and that a common mechanism of action exists. Conceptually, the effects of salt on biopolymer conformational stability, which usually but not always follow the traditional ionic hierarchy, the Hofmeister series, can be viewed as arising in two ways: (a) by a “direct” mechanism, Le., preferential binding of the ionic component to a portion of the macromolecule; (b) by an “indirect” mechanism, i.e., mediated by ion-induced changes in water structure which in turn modulate the energetics of some macromolecular stabilizing interaction such as hydrophobic interaction of apolar moieties. The phenomenology and interpretations of neutral salt effects on biopolymer structures have been reviewed (Tanford, 1968, 1970; von Hippel and Schleich, 1969a,b). The unique molecular interactions (apart from the backbone electrostatic free-energy contribution) which serve to define and stabilize polynucleotide structures are the basebase stacking and hydrogen-bonding interactions and the B I O C H E M I S T R Y , VOL.

11,

NO.

2, 1 9 7 2

137

SMITH

nonbonded interactions of the ribose ring. A number of workers have shown (Donohue and Trueblood, 1960; Haschemeyer and Rich, 1967; Sundaralingam, 1969; Lakshminarayanan and Sasisekharan, 1969a) that the torsional angle which defines the orientation of the nearly planar purine or pyrimidine ring system relative to the furanose ring is crucially dependent on the type and degree of sugar ring puckering. Sundaralingam (1969) has collected evidence which shows that in the solid state the conformation of the furanose ring system in both the monomer and the polymer are remarkably similar (see also, Arnott, 1970). Recently Kreishman and Chan (1971) have extracted the nuclear magnetic resonance (nrnr) parameters from the 220-MHz spectrum of polyuridylic acid. Although they did not point this out, the data show that the ribose conformation of polyuridylic acid is remarkably similar to those of 5’- and 3’-uridylic acid. Six torsional angles (of restricted values) define the conformation of the polynucleotide backbone chain and these are controlled in large part by the prevailing nonbonded interactions in the furanose ring system.’ The importance of restricted rotations about the bonds defining the polynucleotide backbone has support from both conformational energy calculations and X-ray structure determinations, as well as from experimental polymer conformation measurements. The latter has emerged from the lightscattering studies of Felsenfeld and coworkers (Eisenberg and Felsenfeld, 1967; Inners and Felsenfeld, 1970) on synthetic polynucleotides in aqueous solutions at the so-called 8 point (ideal polymer solution conditions). In particular, these workers have shown that the conformational dimensions of the polynucleotide backbone are identical under comparable conditions of stacking for different polynucleotides, thereby suggesting that the nature of the base is relatively unimportant in defining the conformational features of the polynucleotide backbone. Even in the absence of stacking the data are indicative of a highly extended but conformationally restricted polymer chain. Polynucleotide dimensions also appear to be essentially independent of the nature of the salt used to attain ideal solution properties (LiC104US. NaC1) (G. Felsenfeld, personal communication) when compared at the same extent of stdcking. Prestegard and Chan (1969) have interpreted their nmr data to indicate that both the sugar-base torsion angle and the nature of ring puckering of uridine and 5’-UMP are sensitive in a Hofmeister fashion to neutral salt denaturants, but that 3’-UMP and deoxyuridine furanose ring systems were sufficiently rigid to withstand the action of these additives. The sugar-base torsional angles of the latter two still remained sensitive to the presence of salt. In order to improve our understanding of the mechanism of salt-induced polynucleotide denaturation, we initiated a similar series of experiments, confining our study to uridine, uridine 3 ’-phosphate, P-pseudouridine, 0-pseudouridine 3 ’phosphate, and deoxyuridine.2 In view of the extensive data obtained by Prestegard and Chan (1969), only one salt and concentration, 2 M sodium perchlorate was used. Solution temperature is introduced as an additional variable. Our con1

These six angles define the magnitude of rotation about: P-Oj,,

O~,-CS,,C~,-CI,,CI,-C~,,C~,-~~,, and OW-P. 2 Abbreviations used are: U, uridine; dU, deoxyuridine; 3’-UMP, uridine 3 ’-phosphate; p-$, P-pseudouridine, 3 ’-p-$MP, P-pseudouridine 3 ‘-phosphate; poly(rA), polyadenylic acid; poly(rU), polyuridylic acid; UpU, uridylyl-3 ’,5’-uridine; ApA, adenylyl-3 ’ , 5 ’-adenosine; DSS, sodium 3-trimethylsilylpropanesulfonate;TSP, sodium 3-trimethylsilylpropane carboxylate-ds; Dz0, deuterium oxide; HDO, hydrogen deuterium oxide.

138

BIOCHEMISTRY,

VOL.

11,

NO.

2, 1 9 7 2

e t al.

clusions alter the previously established picture of how neutral salts affect nucleotide, nucleoside, and polynucleotide conformations. Experimental Section U, 3’-UMP, 0-$, 3’-P-$MP, and dU were the best grades commercially available and were used as received. Sodium perchlorate (anhydrous) was obtained from G. F. Smith. The two internal references employed in this work, DSS and TSP were obtained from E. Merck, Germany, and Merck, Sharpe and Dohme, Canada, respectively. Nucleosides (and -tides) were made up in D20, or in 2 M NaC104 solutions in DzO, at a concentration of 0.1 M, containing either 0.1 M DSS or TSP as an internal reference. The pD was adjusted to 6.8 i. 0.1 (nucleosides) or 7.2 f 0.1 (nucleotides), and the solutions were lyophilized three times from D20.Spectra were obtained using the Varian HA-100, XL-100, HR-220 (Canadian 220 MHz NMR Center, Ontario Research Foundation, Sheridan Park) nmr spectrometers. Line positions were measured relative to the internal standard following customary procedures. Proton spin-decoupling and INDOR experiments were performed on the XL-100; *lP decoupling was performed on the HA-100 using an NMR Specialties decoupler.

Results and Discussion Spectral Assignment and Interpretation. Complete proton magnetic resonance spectra were taken at 100 and 220 MHz. Assignments were made by references to earlier work (Blackburn et al., 1970; Hruska et a[., 1970a,b), by “-1H decoupling, and by 1H-31P decoupling. The Hzr, resonance of dU, due to the hydrogen cis to the OH at 3’, is assigned to the low-field peaks in accord with the deshielding effect of a cis hydroxyl group (see Hruska et al., 1970b). Spectral analysis was performed using standard methods (Pople et al., 1959) and the computer programmes LAOCOON I I and LAOCNPLT. To avoid the possibility of false solutions spectra were run at two different frequencies and the parameters extracted from spectra at one frequency were used to simulate spectra at the other frequency. Figures 1 and 2 show the experimental and calculated nmr spectra for the ribose region of 3’-/3-$MP and deoxyuridine. Uridine is not shown but may be found in Blackburn et af. (1970). Spectra taken at high temperatures (ca. SOo) and in 2 M salt resembled superficially those in D20alone at ambient temperatures, and thus no problems were encountered in analyzing “perturbed” spectra once the basic spectral parameters were determined. Table I contains the results of our spectral analysis for uracil nucleosides and nucleotides under a variety of conditions. The implications of these results will be discussed in subsequent sections. Sugar-Base Torsion Angle. The sugar-base torsion angle (the dihedral angle formed by the intersection of a pyrimidine base plane with the plane defined by O I ~CI,, , and NI of Ppyrimidine glycosides) typically falls in the range of -10 to - 90” (Sundaralingam, 1969); this conformation is referred to as “anti” in the Donohue-Trueblood convention, and has the 5 and 6 positions of uracil over the ribose ring. The estimation of sugar-base torsional angles for nucleosides and 3’substituted nucleotides by nmr is somewhat more indirect than for 5 ’ nucleotides where phosphate anion electrostatic effects on the base proton resonances offer a convenient measure. One procedure for nucleosides and 3’-substituted nucleo-

PERTURBATIONS OF NUCLEOSIDE CONFORMATIONS

tides is based on the following argument. The uracil Hs resonance sustains a 0.3-ppm downfield shift when bound to a D-ribose moiety in a P-glycosidic linkage (Prestegard and Chan, 1969). This alteration in chemical shift has been rationalized as arising from the combined effect of magnetic anisotropy and electric field gradients due to the proximity of the furanose ether oxygen (010 to the He proton of the uracil moiety. Inspection of molecular models suggests that as the sugar-base torsion angle is made more negative, the He proton swings away from the ribose ring ether oxygen. The calculations of Prestegard and Chan reveal that at a torsional angle of - 60 O both electric field and magnetic anisotropic effects become negligible. Thus if the uforementioned reasoning is correct we must associate upfield shifts in the He pyrimidine resonance with increasingly negative sugar-base torsional angles. Such effects have been demonstrated by Smith et al. (1969) for U, 5'-uMP, UpU, and UpUpU with increasing temperature. The data contained in Table I reveal that perturbation of an aqueous sample of U, o-+, 3'-UMP, 3'-p+MP, or deoxyuridine by either increasing temperature or by the addition of electrolyte denaturant (NaC104) results in an upfield shift in the He resonance, suggesting that perturbation by either temperature or salt increases, in a negative direction, the sugar-base torsional angle. The magnitude of the change cannot be estimated quantitatively. Temperature, however, seems to be more potent in this regard than 2 M NaC104. We must emphasize, however, that conformational calculations and space-filling molecular model studies both indicate that the sugar-base torsional angle is crucially dependent on the mode and nature of the furanose ring buckling, which may also be temperature dependent. Recently it has been pointed out that the chemical shifts of the ribose hydrogens are sensitive to the sugar-base torsion angle in pyrimidine nucleosides (Blackburn et a/., 1970; Dugas et a/., 1971).3This is due to the considerable magnetic anisotropy of the keto group at the 2 position in the pyrimidine base. This effect also depends on the nature of the ribose ring puckering, which can alter the distance between the ribose hydrogens and the 2-keto group. Because the ribose ring spinspin coupling constants (see next section) indicate that the compounds studied here all have essentially the same ribose conformations, any effect of the keto group anisotropy could be clearly inferred from the chemical shift data. The absence of any large changes in ribose hydrogen chemical shifts induced by salt or temperature, plus the comparatively small changes in the shifts of the He resonances, suggests that only a very slight change in the sugar-base torsion angles, within the range of the anti conformation, has occurred. Due to the various approximations involved in this argument, the magnitudes of the changes cannot be estimated. Furanose-Ring Conformation. The furanose-ring conformation is defined by three dihedral angles between the planes containing atoms : CICz"l' and C1C2tH2); C2fC31Hzr and C2C3!H3/ ; C3C4tH3,and C3C4'H4!. Other workers, however, have relied on the dihedral angle specified by planes C1~C2)H1f and C1C2"21, and as we shall see later reliance on this torsional angle as a measure of conformation for furanose-ring structures often leads to erroneous conclusions. 3 For example, the chemical shift differences in parts per million for the corresponding ribose hydrogens of uridine and cu-N-(P-cyanuric acid)-D-ribofuranoside were HI,, -0.20; H2t, -0.37; H3,, -0.14; Hat, $0.18; Hs,, $0.06; Hs,,, +0.08. The HI, resonmce of uridine was 0.20 ppm to higher field from the corresponding resonance in a-N-(Pcyanuric acid)-D-ribofuranoside.

4 75

4 50

3 75

400

4 25 PPM

FIGURE 1: (a) 220-MHz nmr spectrum of M 3'-/?-$MP in DaO, pD 7.2, 23", 0.1 M

the ribose hydrogens of 0.1 DSS. (b) Simulation of the experimental spectrum using the parameters from Table I.

The total number of conformational ring candidates, taking into consideration all possible combinations of bucklings, is 20 (Smith and Jardetzky, 1968). We will, however, confine our interest to the four principal conformations, those which reflect structures commonly found in crystals of nucleosides (and -tides) and in the fiber structures of DNA. Knowledge of the proton-proton coupling constants allows an estimation of the relevant dihedral angle by the use of the Karplus relationship (Karplus, 1959,1963)

where i3 is the dihedral angle between the two coupled protons with coupling constant J , j , and Jo and B are empirical constants. We have selected the constants suggested by Abraham et al. (1962) as best representing the situation for a furanose ring. These constants are Jo = 9.27 (0" < 0 < go"), Jo = 10.36 (90' < 0 < 180°), and B = 0.28 Hz.

630

"

4

50

4 25

400 PPM

2 40

FIGURE 2: (a) 220-MHz nmr spectrum of the ribose hydrogens of 0.1 M deoxyuridine in DIO, pD 6.8,23",0.1 M TSP. (b) Simulation of the

experimental spectrum using the parameters from Table 1. B I O C H E M I S T R Y , VOL.

11,

NO.

2, 1 9 7 2

139

SMITH ~

e t al.

~~~

TABLE I: Chemical Shifts.

and Coupling Constants for Uracil Nucleosides and Nucleotides.

~~

U*

U 23”

U* 23’ NaC104

U 80”

80 a NaC104

P-#

5.901 4.340

5.934 4.369

5.901 4.348

4.221 4.127 3.907 3.803 5.883 7.886

4.238 4.155 3.944 3.835 5.925 7.859

4.8

P-rc.

P-#

70 ” NaC104

dU* 23”

4.668 4.260

4.714 4.254

4.140 4,026 3.863 3.744

4.128 3.993 3.835 3.719

4.135 4.017 3.850 3.737

7.660

7.629

7.627

7.612

6.285 2.379 2.437 4.474 4.060 3.846 3,771 5.888 7.870

4.6

5.0

4.9

5.2

5.1

5.2

5.7

5.0

5.0

5.0

5.0

5.1 3.3 4.6 -12.7

5.6 3.3 5.1 -12.6 b 8.1

5.2 3.2 4.6 -12.7 0.8

5.6 3 .O 5.3 -12.6 0.9

5.2 3.4 4.8 -12.4 0.8

5.6 3.0 5.5 -12.4 0.9

P-JI

30”

30 NaC104

5.934 4.373

4.674 4.279

4.729 4.264

4.222 4.122 3.890 3.798 5.918 7.838

4.240 4.146 3.920 3.825 5.961 7.817

4.141 4.009 3.840 3.726

4.3

5.1

5.2

5.5

5.4 2.9 4.4 -12.7 b 8.1

5.9 3.0 4.7 -12.8 b 8.0

O

70

O

6 1’ 2’ 2“ 3’ 4’ 5’ 5” 5 6

J 1’2’ 1’2“ 2’2’’ 2’3’ 2”3’ 3’4’ 4’5 ’ 4’5” 5’”’ 1’6 56 3’P

220 Frequency of nmr experiment (MHz)

b 8.1

220

220

220

6.5 6.7 -13.9 6.8 3.5 4.0 3.4 5.1 -12.6 b 8.0

100

100

100

100

220

~~

Chemical shifts are expressed relative to internal 0.1 M DSS, except those marked * which are relative to 0.1 M TSP. (To convert data relative to TSP to be relative to DSS, subtract 0.030 ppm.) b Not measurable